News Media Alliance Accuses AI Companies of Violating Copyright Laws, Citing Heavy Use of News Articles to Train Chatbots

News publishers call for "proper permissions and compensations" from AI companies citing overuse of copyrighted material to train language models and generate similar content.

The News Media Alliance (NMA), a trade group representing over 2,200 publishers, has accused major AI companies like OpenAI, Google, and Microsoft of violating copyright laws, alleging that their chatbots replicate content from copyrighted news articles without proper permissions.
NMA research highlights that AI-chatbots heavily rely on news content to train their technology, using it up to 100 times more than other generic data sets when training AI chatbots like ChatGPT and Google's Bard.
NMA claims that these AI systems do not absorb any underlying concepts from the data and produce 'unauthorized derivative works' by responding with paraphrased or outright repetition of memorized portions of the works they were trained on.
The high usage of news material in chatbot training shows the unique value AI companies see in news content, yet these companies are not compensating publishers properly or obtaining necessary permissions for its use, according to the NMA.
The alliance has warned of the potential negative implications if unchecked AI technologies saturate the internet with content, suggesting AI would have nothing left to train on, and it could result in job loss in the journalism industry.